Recognition of polycyclic aromatic hydrocarbons using fluorescence spectrometry combined with bird swarm algorithm...

Area of Science:

• Environmental Chemistry
• Analytical Chemistry
• Computational Chemistry

Background:

• Polycyclic aromatic hydrocarbons (PAHs) are pervasive toxic pollutants in aquatic ecosystems, posing significant risks to ecological and human health.
• Efficient and accurate methods for identifying PAHs are essential for environmental monitoring and risk assessment.

Purpose of the Study:

• To develop and validate a novel machine learning approach for the accurate and efficient identification of typical PAHs mixtures in aqueous solutions.
• To compare the performance of the proposed Bird Swarm Algorithm-optimized Support Vector Machine (BSA-SVM) with other machine learning algorithms for PAH identification.

Main Methods:

• Utilized three-dimensional fluorescence spectroscopy to analyze spectral characteristics of individual PAHs (Acenaphthylene, Fluorene, Naphthalene) and their binary mixtures.
• Developed and applied a Bird Swarm Algorithm-optimized Support Vector Machine (BSA-SVM) for classifying PAH mixtures.
• Evaluated BSA-SVM performance against Particle Swarm Optimization-SVM (PSO-SVM), Genetic Algorithm-SVM (GA-SVM), and standard SVM using spectral data.

Main Results:

• The BSA-SVM model achieved a 100% classification accuracy on the test set, surpassing PSO-SVM, GA-SVM, and SVM.
• BSA-SVM demonstrated the fastest training speed among the optimized SVM algorithms, excluding the original SVM model.
• The proposed method effectively distinguished between individual PAHs and their binary mixtures based on spectral data.

Conclusions:

• The BSA-SVM algorithm is a highly accurate and efficient tool for the qualitative analysis and identification of PAHs mixtures in aquatic environments.
• This machine learning approach offers a promising advancement in environmental monitoring for toxic pollutant detection.